We included 752 E. coli isolates from the collection of The Pennsylvania State University’s E. coli Reference Center in the study (Table 1); this center provides characterization of E. coli isolates submitted from outside sources. Sixty-eight isolates from humans were submitted to the E. coli Reference Center from 9 U.S. states, 45 from Saudi Arabia, 13 from Argentina, 4 from Canada, 3 from Mexico, 3 from Zambia, and 1 from Singapore. Two hundred forty-eight isolates from cattle were submitted from Michigan, 56 from Iowa, 33 from Pennsylvania, 65 from 13 other U.S. states, and 2 from Canada. Fifty-one isolates from turkeys were submitted from 13 U.S. states. Forty-five isolates from chickens were submitted from 10 U.S. states, 2 from Canada, and 2 from India. Twenty-two isolates from swine were submitted from 7 U.S. states, 3 from South Korea, and 1 from India. Seventy-four isolates from nonfood animals were submitted from 20 U.S. states, 5 from Paraguay, and 2 from Hungary. We classified nonfood animals as those not commonly used in food production, including rabbits (19 E. coli isolates), hamsters (8 isolates), deer (7 isolates), horses (7 isolates), dogs (7 isolates), alpacas (5 isolates), okapi (4 isolates), parrots (4 isolates), sheep (4 isolates), antelope (4 isolates), mice (3 isolates), seagulls (2 isolates), a cat (1 isolate), a goat (1 isolate), a llama (1 isolate), a marmoset (1 isolate), a mink (1 isolate), a rat (1 isolate), and a turtle (1 isolate).

Antimicrobial susceptibility testing of all isolates was done with broth microdilution using the PASCO MIC/ID system (Becton, Dickinson and Company, Sparks, MD). Testing was done according to manufacturer’s instructions and according to guidelines developed by the National Committee for Clinical Laboratory Standards (NCCLS) (20). Tested antimicrobials, dilution ranges, and resistance breakpoints are listed in Table 2. Ceftiofur- and cefoxitin-resistant isolates were further examined for production of extended-spectrum--lactamases (ESBLs) with disk diffusion according to NCCLS standards (21).

Isolates were grown at 37°C overnight on veal infusion agar (Becton, Dickinson and Company). A loopful of culture was resuspended in 200 µL of distilled water, incubated at 99°C for 15 min, and centrifuged at 12,000 x g for 2 min. The supernatant was used as a template for amplification of Shiga toxin genes (stx1 and stx2), the intimin gene (eae), and the enterohemolysin A gene (hlyA) through multiplex polymerase chain reaction (PCR) (22). Primers described by Witham et al. (23) and Paton (24) were used for amplification of stx1 and stx2, respectively; those described by Gannon et al. (25) were used for amplification of eae; and those described by Fagan et al. (26) were used for amplification of hlyA. Each 11-µL PCR contained 37.5 ng stx1 primers, 15 ng stx2 primers, 15 ng eae primers, 75 ng hlyA primers, 0.18mM each deoxyribonucleotide, 4.0mM MgCl2, 50mM Tris-HCl (pH 8.3), 275 ng bovine serum albumin, 2% sucrose, 0.1mM Cresol Red (Idaho Technology, Inc., Salt Lake City, UT), and 0.4 U Taq DNA polymerase (PGC Scientifics Corp., Gaithersburg, MD). Reaction contents were cycled as described (11) after which products were electrophoresed in 1% agarose gels at 200 V for 30 min and visualized under ultraviolet light. E. coli O157:H7 (ATCC 43895) was the positive control for all reactions.

Of the isolates in this study, the highest frequencies of antimicrobial-resistant phenotypes were observed for E. coli isolates from humans and turkeys (Figure 1). Fifty-nine percent of isolates from humans were resistant to sulfamethoxazole, 59% to streptomycin, 56% to ampicillin, 56% to tetracycline, 50% to cephalothin, 38% to trimethoprim-sulfamethoxazole, 34% to chloramphenicol, and 18% to amoxicillin-clavulanic acid (Figure 1A). Eighty-four percent of isolates from turkeys were resistant to sulfamethoxazole, followed by 82% to streptomycin, 71% to tetracycline, 49% to ampicillin, 39% to cephalothin, 28% to amoxicillin-clavulanic acid, 24% to gentamicin, and 20% to nalidixic acid (Figure 1B). Nalidixic acid-resistant isolates from turkeys were found to have ciprofloxacin MICs ranging from 0.12 to >8 µg/mL, whereas each of the nalidixic acid-susceptible isolates from these animals were found to have ciprofloxacin MICs of 0.03 µg/mL or less (data not shown).

Resistance profiles among isolates from cattle, chicken, and swine were largely similar to each other (Figure 1). Fifty percent of isolates from cattle were resistant to streptomycin, followed by 47% to tetracycline, 46% to sulfamethoxazole, and 15% to ampicillin (Figure 1C). Seventy-one percent of isolates from chickens were resistant to streptomycin, followed by 63% to tetracycline, 53% to sulfamethoxazole, 20% to gentamicin, 16% to trimethoprim-sulfamethoxazole, and 12% to ampicillin (Figure 1D). Eighty-one percent of isolates from swine were resistant to tetracycline, followed by 62% to streptomycin, 31% to sulfamethoxazole, and 27% to ampicillin (Figure 1E).

Resistance frequencies were lowest for isolates from nonfood animals (Figure 1F); however, 25% were resistant to streptomycin, 20% to sulfamethoxazole, and 18% to tetracycline. Of these streptomycin-, sulfamethoxazole-, and tetracycline-resistant isolates, 76%, 82%, and 67%, respectively, were from companion animals.

Of 174 isolates resistant to ampicillin, 73% were resistant to streptomycin and tetracycline. Of 23 isolates resistant to cefoxitin, 91% were resistant to amoxicillin-clavulanic acid. Each of the five ceftiofur-resistant isolates was resistant to cefoxitin and amoxicillin-clavulanic acid. Based on NCCLS interpretive criteria for confirmatory ESBL testing (21), none of the ceftiofur- or cefoxitin-resistant isolates exhibited phenotypes consistent with ESBL production.

Based on the presence of stx1 and stx2, 26% of the isolates were characterized as STEC. Of these, 89% contained stx1 only, 2% contained stx2 only, and 9% contained both. Eighty-one percent of STEC possessed eae and hlyA, 7% eae only, and 7% hlyA only. Of isolates that were not characterized as STEC, 34% possessed eae and hlyA, 2% eae only, and 24% hlyA only (data not shown).

The highest frequency of STEC was among isolates from cattle, in which 34% were characterized as STEC, followed by 27% of isolates from humans, 14% of isolates from nonfood animals, 12% of isolates from swine, and 6% of isolates from turkeys. None of the isolates from chickens were characterized as STEC.

Of E. coli isolates from cattle, resistance frequencies were generally similar between STEC and other E. coli, respectively, with the exception of ampicillin (26% vs. 8%), chloramphenicol (14% vs. 4%), cephalothin (14% vs. 3%), and trimethoprim-sulfamethoxazole (11% vs. 2%), in which resistance frequencies were noticeably higher (Figure 2A). In contrast, of isolates from humans, resistance frequencies were generally lower among STEC isolates compared with other E. coli (Figure 2B). Specifically, resistance frequencies were lower in STEC compared with other E. coli, respectively, for ampicillin (14% vs. 71%), chloramphenicol (5% vs. 44%), sulfamethoxazole (30% vs. 68%), cephalothin (11% vs. 64%), tetracycline (32% vs. 63%), trimethoprim-sulfamethoxazole (8% vs. 48%), amoxicillin-clavulanic acid (5% vs. 22%), and streptomycin (32% vs. 67%).